I
A
E
S
I
n
t
e
r
n
at
io
n
al
Jou
r
n
al
of
A
r
t
if
ic
ia
l
I
n
t
e
ll
ig
e
n
c
e
(
I
J
-
AI
)
V
ol
.
13
, N
o.
4
,
D
e
c
e
m
be
r
2024
, pp.
4106
~
4112
I
S
S
N
:
2252
-
8938
,
D
O
I
:
10.11591/
ij
a
i.
v
13
.i
4
.pp
4106
-
4112
4106
Jou
r
n
al
h
om
e
page
:
ht
tp
:
//
ij
ai
.
ia
e
s
c
or
e
.c
om
D
e
t
e
c
t
i
n
g f
r
au
d
u
l
e
n
t
f
i
n
an
c
i
al
st
at
e
m
e
n
t
u
n
d
e
r
i
m
b
a
l
an
c
e
d
d
at
a
u
si
n
g
n
e
u
r
al
n
e
t
w
or
k
H
e
n
d
r
a T
j
ah
yad
i
, Y
os
u
a E
f
r
ai
m
Y
ou
n
g
S
t
udy P
r
ogr
a
m
of
I
nf
or
m
a
t
i
c
s
,
F
a
c
ul
t
y of
C
om
put
e
r
S
c
i
e
nc
e
,
U
ni
ve
r
s
i
t
a
s
P
e
l
i
t
a
H
a
r
a
pa
n, J
a
ka
r
t
a
, I
ndone
s
i
a
A
r
t
ic
le
I
n
f
o
A
B
S
T
R
A
C
T
A
r
ti
c
le
h
is
to
r
y
:
R
e
c
e
iv
e
d
D
e
c
29
,
2023
R
e
vi
s
e
d
A
pr
19
,
2024
A
c
c
e
pt
e
d
J
un
8
,
2024
In
th
is
p
ape
r
a
no
ve
l
app
ro
ac
h
for
d
et
ec
tin
g
fr
aud
ul
ent
fin
an
cia
l
st
at
em
ent
s
by
emp
loy
in
g
a
c
om
bin
at
ion
of
neu
ra
l
n
etw
or
ks
a
nd
syn
th
eti
c
m
ino
ri
ty
ove
r
-
sam
pli
ng
t
ec
hni
qu
e
(S
MOT
E)
i
s
in
tr
odu
ce
d.
T
his
app
ro
ac
h
is
d
es
ign
ed
t
o
t
ack
le
the
p
ro
ble
m
of
imb
al
anc
ed
d
at
ase
ts
p
re
val
en
t
in
f
ra
udu
le
nt
c
ase
s,
w
hi
c
h
if
l
eft
una
ddr
es
sed
wil
l
hi
nd
er
t
he
m
od
el
t
o
ac
cur
at
el
y
id
ent
if
y
fr
aud
.
Th
re
e
n
eur
al
net
wor
k
mo
de
ls
, ea
ch
r
ep
res
en
tin
g
di
ff
ere
nt
fra
ud
p
re
dic
to
rs
a
s
th
e
in
pu
t
l
aye
r
:
28
inp
ut
s
r
aw
fi
nan
ci
al
da
ta
;
1
4
i
np
uts
f
ina
n
c
ial
r
ati
os
da
ta
;
a
nd
42
in
put
s
com
bin
at
ion
b
oth
r
aw
fi
na
nci
al
an
d
fin
an
ci
al
ra
tio
s
dat
a
ar
e
d
ev
elo
ped
.
Exp
eri
me
nta
l
v
ali
da
tio
n
us
in
g
es
ta
bli
sh
ed
r
ese
ar
ch
d
at
ase
ts
i
s
c
ond
u
cte
d
to
ass
ess
th
e
per
fo
rma
nc
e
o
f
the
pr
op
ose
d
met
ho
d.
Per
fo
rma
nc
e
m
et
ric
s,
nam
ely
are
a und
er
t
he
cu
rv
e
(AU
C)
, pre
ci
sio
n,
a
nd se
nsi
ti
vit
y,
a
re
us
ed
f
or
ev
al
uat
ion
,
com
par
in
g
t
he
p
ro
pos
ed
mo
del
ag
ai
nst
ex
ist
in
g
b
enc
hm
ark
mo
de
ls
f
o
und
in
lit
era
tu
re.
Re
su
lts
in
di
cat
e
t
hat
th
e
p
ro
pos
ed
mod
el
ac
hie
ve
s
a
n
A
UC
s
cor
e
o
f
70.
6%
and
a
pr
eci
si
on
sco
r
e
of
2.
89%
,
i
n
c
om
par
ab
le
to
th
e
e
xis
ti
ng
mod
els
,
wit
h
a
s
ens
it
ivi
ty
sc
or
e
o
f
83%
o
utp
er
fo
rmi
ng
al
l
cou
nt
erp
ar
ts.
T
h
e
h
igh
sen
sit
iv
ity
rat
e
of
t
he
p
ro
pos
ed
m
od
el
u
nde
rs
cor
es
i
ts
p
rac
ti
ca
l
uti
l
ity
f
or
aud
ito
rs
a
nd
r
egu
la
tor
s,
a
s
it
m
ini
mi
zes
the
r
is
k
of
f
al
s
e
ne
gat
iv
es,
t
her
eby
enh
anc
in
g c
on
fid
en
ce
in
f
rau
d
det
ec
tio
n.
K
e
y
w
o
r
d
s
:
F
r
a
udul
e
nt
f
in
a
nc
ia
l
s
ta
te
m
e
nt
s
M
a
c
hi
ne
l
e
a
r
ni
ng
N
e
ur
a
l
ne
twor
k
S
upe
r
vi
s
e
d l
e
a
r
ni
ng
S
ynt
he
ti
c
m
in
or
it
y
ove
r
-
s
a
m
pl
in
g t
e
c
hni
que
This is an
open
acce
ss artic
le unde
r the
CC BY
-
SA
license.
C
or
r
e
s
pon
di
n
g A
u
th
or
:
H
e
ndr
a
T
ja
hya
di
S
tu
dy P
r
ogr
a
m
of
I
nf
or
m
a
ti
c
s
, F
a
c
ul
ty
of
C
om
put
e
r
S
c
ie
nc
e
, U
ni
ve
r
s
it
a
s
P
e
li
ta
H
a
r
a
pa
n
J
a
ka
r
ta
, I
ndone
s
ia
E
m
a
il
:
he
ndr
a
.t
ja
hya
di
@
uph.e
du
1.
I
N
T
R
O
D
U
C
T
I
O
N
F
in
a
nc
ia
l
s
ta
te
m
e
nt
m
is
s
ta
te
m
e
nt
s
m
a
y a
r
is
e
f
r
om
e
it
he
r
f
r
a
ud
or
e
r
r
or
,
a
s
s
ta
te
d by the
I
nt
e
r
na
ti
ona
l
F
e
de
r
a
ti
on
of
A
c
c
ount
a
nt
s
[
1]
.
I
t
is
th
e
a
udi
to
r
’
s
r
e
s
pon
s
ib
il
it
y
to
pr
ovi
de
r
e
a
s
ona
bl
e
a
s
s
ur
a
nc
e
th
a
t
th
e
f
in
a
nc
ia
l
s
ta
te
m
e
nt
s
a
r
e
f
r
e
e
f
r
om
m
a
te
r
ia
l
m
is
s
ta
te
m
e
nt
.
M
is
le
a
di
ng
f
in
a
nc
ia
l
s
ta
te
m
e
nt
s
c
a
n
in
c
ur
s
ig
ni
f
ic
a
nt
c
os
ts
,
e
s
p
e
c
ia
ll
y
f
or
in
ve
s
to
r
s
,
r
e
gul
a
to
r
s
,
a
nd
s
oc
ie
ty
a
t
la
r
ge
,
a
s
de
m
ons
tr
a
te
d
in
th
e
E
nr
on
s
c
a
nda
l
—
one
of
th
e
m
os
t
not
a
b
le
a
udi
t
a
nd a
c
c
ount
in
g s
c
a
nda
ls
i
n hi
s
to
r
y a
nd l
i
te
r
a
tu
r
e
[
2
]
–
[
4
]
. I
t
be
ga
n w
he
n E
nr
on s
hoc
ke
d
th
e
publi
c
by r
e
por
ti
ng a
$638
m
il
li
on
l
os
s
. T
hi
s
c
a
s
e
i
m
pl
ic
a
te
d i
ts
a
udi
to
r
, A
r
th
ur
A
nde
r
s
e
n, w
hi
c
h
f
a
il
e
d t
o
de
te
c
t
th
e
m
is
s
ta
te
m
e
nt
a
nd
e
nga
g
e
d
in
doc
um
e
nt
s
hr
e
ddi
ng
r
e
la
te
d
to
E
nr
on
a
udi
ts
.
T
hi
s
hi
ghl
ig
ht
s
th
e
di
f
f
ic
ul
ty
i
n de
te
c
ti
ng a
c
c
ount
in
g m
is
s
ta
te
m
e
nt
s
.
D
e
te
c
ti
ng
a
c
c
ount
in
g
m
is
s
t
a
te
m
e
nt
s
c
a
n
be
c
ha
ll
e
ngi
ng
due
to
s
e
ve
r
a
l
r
e
a
s
ons
s
uc
h
a
s
th
e
c
om
pl
e
xi
ty
of
f
in
a
nc
ia
l
tr
a
ns
a
c
ti
ons
,
s
ophi
s
ti
c
a
te
d
f
r
a
ud
s
c
he
m
e
s
,
va
s
t
a
m
ount
of
da
ta
,
a
nd
hum
a
n
e
r
r
or
a
nd
bi
a
s
.
T
he
s
e
c
ha
ll
e
nge
s
unde
r
s
c
or
e
th
e
ne
e
d
f
or
in
nova
ti
ve
a
ppr
oa
c
h
s
uc
h
a
s
da
ta
a
na
ly
ti
c
s
a
nd
m
a
c
hi
ne
le
a
r
ni
ng
in
a
udi
ti
ng
[
3]
–
[
5]
.
T
he
s
e
te
c
hnol
ogi
e
s
of
f
e
r
th
e
pot
e
nt
ia
l
to
e
nha
nc
e
a
udi
t
e
f
f
e
c
ti
ve
ne
s
s
,
im
pr
ove
r
is
k
a
s
s
e
s
s
m
e
nt
,
a
nd
m
it
ig
a
te
t
he
i
m
pa
c
t
of
huma
n l
im
it
a
ti
ons
on a
udi
t
qua
li
ty
.
A
lt
hough
da
ta
a
n
a
ly
ti
c
s
a
nd
m
a
c
hi
ne
le
a
r
ni
ng
a
r
e
e
xpe
c
te
d
to
de
m
ons
tr
a
te
a
s
upe
r
io
r
m
e
th
od,
th
e
y
a
r
e
s
e
ld
om
to
us
e
on
p
e
r
f
or
m
in
g
a
udi
t
pr
oc
e
dur
e
s
.
It
is
r
e
la
ti
ve
ly
unknown
w
he
th
e
r
us
a
ge
of
da
ta
a
na
ly
ti
c
s
a
nd
Evaluation Warning : The document was created with Spire.PDF for Python.
I
nt
J
A
r
ti
f
I
nt
e
ll
I
S
S
N
:
2252
-
8938
D
e
te
c
ti
ng f
r
audule
nt
f
in
anc
ia
l
s
ta
te
m
e
nt
und
e
r
i
m
bal
anc
e
d dat
a us
in
g ne
ur
al
ne
tw
o
r
k
(
Y
os
ua E
fr
ai
m
Y
oung
)
4107
m
a
c
hi
ne
le
a
r
ni
ng
a
r
e
in
de
e
d
tr
a
ns
f
or
m
a
ti
ona
l
f
or
th
e
a
udi
t
[
6]
.
V
a
r
io
us
r
e
s
e
a
r
c
h
ha
s
be
e
n
c
onduc
t
e
d
in
s
e
a
r
c
hi
ng
of
f
r
a
udul
e
nt
f
in
a
nc
ia
l
s
ta
te
m
e
nt
de
te
c
ti
on,
in
c
lu
di
ng
ut
il
iz
a
ti
on
of
s
upe
r
vi
s
e
d
le
a
r
ni
ng
a
nd
uns
upe
r
vi
s
e
d l
e
a
r
ni
ng. S
upe
r
vi
s
e
d l
e
a
r
ni
ng i
s
us
e
d i
nc
lu
di
ng va
r
io
us
m
ode
ls
s
uc
h a
s
ne
ur
a
l
ne
twor
k
[
7]
–
[
11]
,
ge
ne
ti
c
a
lg
or
it
hm
[
12]
,
de
c
is
io
n
tr
e
e
(
D
T
)
[
8]
–
[
10
]
,
[
13]
,
B
a
ye
s
ia
n
ne
twor
k
[
8]
,
[
9]
,
s
uppor
t
ve
c
to
r
m
a
c
hi
ne
s
(
S
V
M
)
[
8]
, [
13]
–
[
15]
, a
nd
lo
gi
s
ti
c
r
e
gr
e
s
s
io
n
(
L
R
)
[
16]
. U
ns
upe
r
vi
s
e
d l
e
a
r
ni
ng i
m
pl
e
m
e
nt
a
ti
on us
e
a
lg
or
it
hm
s
uc
h
a
s
s
e
lf
-
or
ga
ni
z
in
g
m
a
p
[
17]
,
[
18]
a
nd
k
-
m
e
a
ns
c
lu
s
te
r
in
g
[
17]
.
O
ne
s
ig
ni
f
ic
a
nt
obs
ta
c
le
in
m
a
c
hi
ne
le
a
r
ni
ng
is
th
e
im
ba
la
nc
e
d
da
ta
c
ha
ll
e
nge
,
w
he
r
e
une
qua
l
c
la
s
s
r
e
pr
e
s
e
nt
a
ti
on
le
a
ds
to
in
a
c
c
ur
a
te
de
te
c
ti
on,
w
it
h
m
a
jo
r
it
y
c
la
s
s
e
s
ove
r
s
ha
dow
in
g
m
in
or
it
ie
s
.
P
ubl
ic
ly
a
v
a
il
a
bl
e
f
in
a
nc
ia
l
s
ta
te
m
e
nt
s
of
te
n
e
xhi
bi
t
s
e
ve
r
e
im
ba
la
nc
e
due
to
th
e
r
a
r
it
y
of
f
r
a
udul
e
nt
in
s
ta
nc
e
s
c
om
pa
r
e
d
to
non
-
f
r
a
udul
e
nt
one
s
.
T
he
r
e
f
or
e
,
it
is
c
r
uc
ia
l
f
or
m
ode
ls
t
o a
ddr
e
s
s
t
hi
s
i
m
ba
la
nc
e
.
T
hi
s
r
e
s
e
a
r
c
h a
im
s
t
o d
e
ve
lo
p a
m
ode
l
f
or
pr
e
di
c
ti
ng f
r
a
udul
e
nt
f
in
a
nc
ia
l
s
ta
te
m
e
nt
s
f
r
om
r
e
a
l
publ
ic
da
ta
s
e
ts
a
nd t
o t
a
c
kl
e
t
he
i
m
ba
la
nc
e
d da
ta
i
s
s
ue
. T
hr
e
e
ne
ur
a
l
n
e
twor
ks
m
ode
ls
w
it
h di
f
f
e
r
e
nt
t
ype
s
of
in
put
s
,
na
m
e
ly
r
a
w
f
in
a
nc
ia
l
d
a
ta
,
f
in
a
nc
ia
l
r
a
ti
os
da
ta
,
a
nd
a
c
om
bi
na
ti
on
th
e
r
e
of
,
c
om
bi
ne
d
w
it
h
synthetic
m
in
or
it
y
ove
r
-
s
a
m
pl
in
g
te
c
hni
que
(
S
M
O
T
E
)
,
a
r
e
pr
opos
e
d
in
th
is
s
tu
dy.
T
he
r
e
s
t
of
th
i
s
pa
p
e
r
is
or
ga
ni
z
e
d
a
s
f
ol
lo
w
s
:
i
n s
e
c
ti
on 2, w
e
out
li
ne
pr
e
vi
ous
e
f
f
or
ts
by r
e
s
e
a
r
c
he
r
s
t
o de
t
e
c
t
f
r
a
udul
e
nc
e
i
n f
in
a
nc
ia
l
s
ta
te
m
e
nt
, bot
h u
s
in
g
c
om
m
onl
y
ba
la
nc
e
d
s
im
ul
a
ti
on
da
ta
a
nd
im
ba
la
nc
e
d
r
e
a
l
d
a
t
a
.
S
e
c
ti
on
3
de
t
a
il
s
th
e
m
e
th
od
w
e
pr
opos
e
,
ut
il
iz
in
g
a
c
om
bi
na
ti
on
of
ne
ur
a
l
ne
twor
ks
a
nd
S
M
O
T
E
to
de
te
c
t
f
r
a
udul
e
nc
e
in
hi
ghl
y
im
ba
la
nc
e
d
r
e
a
l
da
t
a
.
I
n
s
e
c
ti
on
4,
th
e
e
xpe
r
im
e
nt
a
l
r
e
s
ul
ts
a
r
e
pr
e
s
e
nt
e
d
a
nd
c
om
pa
r
e
d
w
it
h
th
os
e
of
pr
e
vi
ou
s
r
e
s
e
a
r
c
he
r
s
.
F
in
a
ll
y,
in
s
e
c
ti
on 5, we
c
onc
lu
d
e
w
it
h a
s
um
m
a
r
y of
our
f
in
di
ngs
.
2.
L
I
T
E
R
A
T
U
R
E
R
E
V
I
E
W
T
he
e
xi
s
ti
ng
li
te
r
a
tu
r
e
ha
s
f
oc
us
e
d
c
ons
id
e
r
a
bl
e
a
tt
e
nt
io
n
on
f
in
a
nc
ia
l
da
ta
a
s
c
r
uc
ia
l
in
di
c
a
to
r
s
of
f
r
a
ud,
e
nc
om
pa
s
s
in
g
bot
h
r
a
w
f
in
a
nc
ia
l
d
a
ta
a
nd
f
in
a
n
c
ia
l
r
a
t
io
s
.
A
s
f
unda
m
e
nt
a
l
c
om
pone
nt
s
of
f
in
a
nc
ia
l
s
ta
te
m
e
nt
s
,
f
in
a
nc
ia
l
d
a
ta
ha
ve
th
e
pot
e
nt
ia
l
to
in
di
c
a
te
f
r
a
ud
r
is
k.
F
or
e
xa
m
pl
e
,
a
li
qui
di
ty
r
a
ti
o
de
r
iv
e
d
di
r
e
c
tl
y
f
r
om
r
a
w
f
in
a
nc
ia
l
da
ta
c
oul
d
s
e
r
ve
a
s
a
n
e
f
f
e
c
ti
ve
m
e
a
s
ur
e
of
a
c
om
pa
ny'
s
f
in
a
nc
ia
l
pr
e
s
s
ur
e
.
T
hi
s
unde
r
s
c
or
e
s
th
e
s
upe
r
io
r
it
y
of
c
e
r
ta
in
r
a
ti
os
ove
r
ot
he
r
s
[
19]
,
pa
r
ti
c
ul
a
r
ly
th
os
e
f
in
a
nc
ia
l
da
ta
poi
nt
s
c
lo
s
e
ly
li
nke
d
to
th
e
f
r
a
ud
tr
ia
ngl
e
th
e
or
y.
B
y
le
v
e
r
a
gi
ng
f
in
a
nc
ia
l
d
a
ta
,
s
e
ve
r
a
l
a
p
pr
oa
c
he
s
ut
il
iz
in
g
m
a
c
hi
ne
le
a
r
ni
ng
a
nd
d
a
ta
m
in
in
g t
o de
te
c
t
f
r
a
udul
e
nt
f
in
a
nc
ia
l
s
ta
te
m
e
nt
a
r
e
f
ound in t
he
l
it
e
r
a
tu
r
e
.
G
r
e
e
n
a
nd
C
hoi
[
7]
de
m
on
s
tr
a
te
d
th
e
pot
e
nt
ia
l
of
n
e
ur
a
l
ne
two
r
k
a
ppl
ic
a
ti
ons
in
f
r
a
ud
in
ve
s
ti
ga
ti
on
a
nd
ut
il
iz
e
d
it
a
s
a
de
te
c
ti
on
to
ol
,
e
m
pl
oyi
ng
172
s
a
m
pl
e
s
,
w
it
h
86
s
a
m
pl
e
s
f
or
bot
h
f
r
a
udul
e
nt
a
nd
non
-
f
r
a
udul
e
nt
c
a
s
e
s
.
T
he
m
ode
l
a
c
hi
e
ve
d
a
n
a
c
c
ur
a
c
y
r
a
te
of
74%
.
K
ot
s
ia
nt
is
e
t
al
.
[
8
]
c
onduc
te
d
e
xpe
r
im
e
nt
s
on
DT
,
a
r
ti
f
ic
ia
l
ne
ur
a
l
ne
twor
k (
ANNs
)
,
B
a
ye
s
i
a
n ne
twor
ks
, r
ul
e
l
e
a
r
ne
r
s
,
ne
a
r
e
s
t
ne
ig
hbor
s
,
a
nd
S
V
M
.
T
hi
s
s
tu
dy
de
m
ons
tr
a
te
d
th
a
t
DT
out
pe
r
f
or
m
e
d
ot
he
r
m
ode
ls
w
it
h
9
1.2%
a
c
c
ur
a
c
y
u
s
in
g
a
ba
la
nc
e
d
da
ta
s
e
t
of
164
G
r
e
e
k
c
om
pa
ni
e
s
li
s
te
d
on
th
e
A
th
e
ns
s
to
c
k
e
xc
h
a
nge
,
c
om
pr
is
i
ng
41
f
r
a
udul
e
nt
a
nd
123
non
-
f
r
a
udul
e
nt
c
a
s
e
s
.
I
n
a
s
im
il
a
r
w
or
ks
,
K
ir
kos
e
t
al
.
[
9]
c
onduc
te
d
e
xpe
r
im
e
nt
s
u
s
in
g
DT
,
ne
ur
a
l
ne
twor
k
,
a
nd
B
a
ye
s
ia
n
be
li
e
f
ne
twor
ks
, r
e
ve
a
li
ng t
ha
t
B
a
y
e
s
ia
n
be
li
e
f
ne
twor
k
s
out
pe
r
f
or
m
e
d ot
he
r
s
w
it
h 90.3%
a
c
c
ur
a
c
y us
in
g
a
ba
la
nc
e
d
da
ta
s
e
t
of
76 G
r
e
e
k m
a
nuf
a
c
tu
r
in
g c
om
pa
ni
e
s
, i
nc
lu
di
ng 38 f
r
a
udul
e
nt
a
nd 38 non
-
f
r
a
udul
e
nt
c
a
s
e
s
.
C
e
c
c
hi
ni
e
t
a
l.
[
1
4]
u
s
in
g
S
V
M
,
a
c
c
ur
a
te
ly
id
e
nt
if
i
e
d
80%
of
f
r
a
udul
e
nt
c
a
s
e
s
a
n
d
9
0.6%
of
non
-
f
r
a
udul
e
nt
c
a
s
e
s
f
r
om
a
da
t
a
s
e
t
c
om
pr
is
in
g
6
,427
non
-
f
r
a
u
dul
e
nt
a
n
d
2
05
f
r
a
udul
e
nt
s
a
m
pl
e
s
.
T
hi
s
s
tu
d
y
w
a
s
c
on
s
id
e
r
e
d
a
pi
one
e
r
in
g
w
or
k
in
t
he
f
i
e
ld
.
D
e
c
how
e
t
al
.
[
16]
pr
e
s
e
nt
e
d
a
n
a
l
te
r
n
a
ti
ve
m
e
th
od
u
s
in
g
LR
w
it
h
f
in
a
n
c
ia
l
r
a
ti
os
t
o
de
te
c
t
f
r
a
ud
ul
e
nt
f
in
a
nc
ia
l
s
ta
t
e
m
e
nt
s
,
s
i
gna
li
ng
th
e
li
k
e
li
ho
od
of
m
is
s
ta
te
m
e
nt
.
P
e
r
ol
s
[
10]
,
w
it
h
a
la
r
g
e
r
d
a
ta
s
e
t
of
15,93
4
no
n
-
f
r
a
udul
e
nt
a
nd
51
f
r
a
ud
ul
e
nt
c
a
s
e
s
,
d
e
m
on
s
tr
a
te
d
th
a
t
LR
a
n
d
S
V
M
o
ut
pe
r
f
or
m
e
d
n
e
ur
a
l
ne
t
w
or
k
,
b
a
g
gi
ng,
C
.4
5,
a
nd
s
ta
c
ki
n
g
a
lg
or
it
hm
.
T
he
s
e
f
in
di
ng
s
w
e
r
e
c
on
s
i
s
te
nt
w
it
h
th
os
e
of
Y
a
o
e
t
a
l.
[
20]
,
w
hi
c
h
s
ho
w
e
d
th
a
t
S
V
M
ha
d
th
e
hi
g
he
s
t
a
c
c
ur
a
c
y
a
m
ong
v
a
r
io
u
s
c
la
s
s
if
i
c
a
t
io
n
m
e
th
od
s
.
R
a
nd
ha
w
a
e
t
al
.
[
21]
in
ve
s
ti
g
a
te
d
th
e
e
f
f
e
c
ti
ve
n
e
s
s
of
s
in
g
le
a
nd
h
ybr
id
m
e
t
hod
s
,
e
m
p
lo
yi
ng
unde
r
-
s
a
m
pl
i
ng
to
de
t
e
c
t
c
r
e
di
t
c
a
r
d
f
r
a
ud.
T
he
ir
s
tu
dy
r
e
ve
a
le
d
th
a
t
c
om
b
in
in
g
A
d
a
B
oo
s
t
a
nd
m
a
j
or
it
y
v
ot
in
g
m
e
th
od
s
yi
e
ld
e
d
th
e
b
e
s
t
r
e
s
ul
t
s
.
B
a
o
e
t
al
.
[
15]
e
xt
e
nd
e
d
th
is
r
e
s
e
a
r
c
h
by
us
in
g
a
la
r
ge
p
ubl
ic
da
ta
s
e
t
a
nd
c
om
pa
r
e
d
th
e
r
e
s
ul
t
s
of
r
e
-
im
pl
e
m
e
nt
in
g
th
e
m
ode
l
s
pr
opo
s
e
d
in
[
14]
,
[
16]
w
it
h
a
ne
w
s
ta
t
e
-
of
-
th
e
-
a
r
t
m
ode
l
us
in
g
R
U
S
B
oo
s
t
.
T
he
pr
opo
s
e
d
m
e
th
od
out
p
e
r
f
or
m
e
d
th
e
pr
e
vi
ous
m
ode
l
s
w
it
h
a
n
a
r
e
a
unde
r
c
ur
v
e
(
A
U
C
)
of
72.5%
a
nd
s
e
n
s
it
i
vi
ty
a
n
d
pr
e
c
is
io
n
of
4.88%
a
nd
4.48%
,
r
e
s
p
e
c
ti
ve
l
y.
H
o
a
ng
e
t
al
.
[
22]
e
m
pl
oye
d
X
G
B
oos
t
a
nd
f
-
X
G
B
oo
s
t
on
th
e
d
a
ta
s
e
t
us
e
d
in
[
15]
,
r
e
s
ul
ti
ng
in
A
U
C
s
c
or
e
s
of
68.
9%
a
nd
69.3%
,
a
nd
pr
e
c
i
s
io
n
a
nd
s
e
n
s
it
iv
i
ty
of
3.
56%
,
5%
,
a
nd
3
.36%
,
a
nd
4.
22%
,
r
e
s
p
e
c
ti
v
e
ly
.
A
s
ht
i
a
ni
a
nd
R
a
a
he
m
i
[
3]
f
o
und
th
a
t
a
s
in
gl
e
m
ode
l
out
pe
r
f
or
m
e
d both e
n
s
e
m
b
le
a
nd
hybr
id
a
ppr
oa
c
he
s
. T
h
e
y
hi
ghl
i
ght
e
d T
e
m
p
one
r
a
s
e
t
al
.
[
23]
a
p
pr
oa
c
h
o
f
e
m
pl
oyi
ng
a
de
e
p d
e
n
s
e
m
ul
ti
l
a
ye
r
pe
r
c
e
pt
r
on
,
a
c
hi
e
vi
ng
a
n
a
c
c
ur
a
c
y
of
93
.7%
u
s
in
g
a
da
t
a
s
e
t
of
164
G
r
e
e
k
c
om
pa
ni
e
s
. C
r
a
ja
e
t
al
.
[
24]
us
in
g a
t
e
xt
m
i
ni
ng a
p
pr
oa
c
h t
o de
t
e
c
t
f
r
a
ud
ul
e
nt
f
in
a
nc
i
a
l
s
t
a
te
m
e
nt
s
f
r
om
a
nn
ua
l
r
e
por
ts
,
d
e
m
on
s
tr
a
t
e
d
th
e
e
f
f
e
c
ti
v
e
ne
s
s
a
nd
pr
e
f
e
r
e
n
c
e
f
or
A
N
N
,
e
m
pha
s
iz
i
ng
t
he
ir
a
bi
li
ty
to
c
a
p
tu
r
e
c
om
pl
e
x
r
e
la
ti
on
s
hi
ps
a
m
o
ng
v
a
r
ia
b
le
s
.
I
n
s
pi
r
e
d
b
y
th
e
e
f
f
e
c
ti
ve
n
e
s
s
of
n
e
ur
a
l
ne
t
w
or
ks
,
th
i
s
s
tu
dy
pr
opo
s
e
s
a
n
a
ppr
o
a
c
h
c
om
bi
ni
ng n
e
ur
a
l
ne
tw
or
ks
a
nd
S
M
O
T
E
t
o
de
t
e
c
t
f
r
a
udu
le
nt
f
i
n
a
nc
i
a
l
s
ta
t
e
m
e
nt
s
i
n a
n
i
m
ba
la
n
c
e
d
da
t
a
s
e
t.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
2252
-
8938
I
nt
J
A
r
ti
f
I
nt
e
ll
,
V
ol
.
13
, N
o.
4
,
D
e
c
e
m
be
r
2024
:
4106
-
4112
4108
3.
M
E
T
H
O
D
T
he
de
te
c
ti
on
m
ode
l
f
or
f
r
a
udul
e
nt
f
in
a
nc
ia
l
s
ta
te
m
e
nt
pr
opos
e
d
in
th
is
s
tu
dy
ut
il
iz
e
s
a
c
om
bi
na
ti
on
of
ne
ur
a
l
ne
twor
ks
a
nd
S
M
O
T
E
.
I
ni
ti
a
ll
y,
a
s
e
ve
r
e
ly
im
ba
la
nc
e
d
publ
ic
da
ta
s
e
t
c
ont
a
in
in
g
r
e
a
l
f
in
a
nc
ia
l
s
ta
te
m
e
nt
s
is
a
c
qui
r
e
d.
S
ubs
e
qu
e
nt
ly
,
th
e
da
ta
s
e
t
u
nde
r
goe
s
pr
e
pr
oc
e
s
s
in
g
to
a
ddr
e
s
s
th
e
im
ba
la
nc
e
d
da
ta
s
e
t
us
in
g
S
M
O
T
E
,
w
hi
c
h
ge
ne
r
a
te
s
s
ynt
he
ti
c
s
a
m
pl
e
s
f
or
th
e
m
in
or
it
y
c
la
s
s
.
T
he
pr
e
pr
oc
e
s
s
e
d
da
ta
i
s
th
e
n
us
e
d
f
or
tr
a
in
in
g
a
nd
e
xpe
r
im
e
nt
a
ti
on
on
th
e
p
r
opos
e
d
ne
twor
k
m
ode
ls
.
F
in
a
ll
y,
th
e
r
e
s
ul
ts
obt
a
in
e
d
f
r
om
e
m
pl
oyi
ng
ne
ur
a
l
ne
twor
ks
a
r
e
c
om
pa
r
e
d
to
th
os
e
a
c
hi
e
ve
d
by
th
e
s
ta
te
-
of
-
th
e
-
a
r
t
a
lg
o
r
it
hm
pr
opos
e
d
in
p
r
e
vi
ous
li
te
r
a
tu
r
e
. T
he
ove
r
a
ll
pr
oc
e
s
s
w
or
kf
lo
w
i
s
i
ll
us
tr
a
te
d i
n F
ig
ur
e
1.
F
ig
ur
e
1. P
r
opos
e
d f
r
a
ud de
te
c
ti
on w
or
kf
lo
w
3.1.
D
at
a an
d
v
ar
ia
b
le
s
T
he
da
ta
s
e
t
is
r
e
tr
ie
ve
d
f
r
om
pr
e
vi
ous
r
e
s
e
a
r
c
h
by
B
a
o
e
t
a
l.
[
15]
,
c
om
pr
is
in
g
146,045
r
e
c
or
ds
c
ol
le
c
te
d
f
r
om
th
e
C
O
M
P
U
S
T
A
T
da
ta
ba
s
e
,
c
ov
e
r
in
g
a
ll
publ
ic
ly
li
s
te
d
U
.S
.
f
ir
m
s
f
r
om
1990
to
2014.
I
t
in
c
lu
de
s
42
f
e
a
tu
r
e
s
,
c
on
s
is
ti
ng
of
28
r
a
w
f
in
a
nc
ia
l
da
ta
de
r
iv
e
d
f
r
om
r
e
s
e
a
r
c
h
by
C
e
c
c
hi
ni
e
t
al
.
[
14]
a
nd
14
f
in
a
nc
ia
l
r
a
ti
os
a
s
r
e
s
e
a
r
c
he
d
by
D
e
c
how
e
t
al
.
[
16
]
.
F
ol
lo
w
in
g
th
e
pr
e
vi
ous
s
tu
dy,
th
e
tr
a
in
in
g
da
ta
s
e
t
s
pa
ns
f
r
om
1991 to t
he
t
e
s
t
ye
a
r
, w
it
h a
t
w
o
-
ye
a
r
ga
p.
S
e
r
ia
l
f
r
a
ud,
de
f
in
e
d
a
s
f
r
a
udul
e
nt
c
a
s
e
s
s
pa
nni
ng
m
ul
ti
pl
e
ye
a
r
s
,
is
pr
e
s
e
nt
in
th
e
da
ta
s
e
t.
T
he
im
pa
c
t
of
s
e
r
ia
l
f
r
a
ud
is
th
a
t
it
c
a
n
in
f
la
te
th
e
m
ode
l'
s
pe
r
f
or
m
a
nc
e
,
a
s
th
e
s
a
m
e
f
r
a
ud
c
a
s
e
m
a
y
be
in
c
lu
de
d
in
bot
h
th
e
tr
a
in
a
nd t
e
s
t
da
ta
[
15]
. T
he
r
e
f
or
e
,
to
pr
e
ve
nt
ove
r
s
ta
te
d r
e
s
ul
ts
a
nd be
nc
hm
a
r
k a
ga
in
s
t
pr
e
vi
ous
l
it
e
r
a
tu
r
e
, t
he
da
ta
s
e
t
is
pr
e
pr
oc
e
s
s
e
d by r
e
c
odi
ng a
ll
s
e
r
ia
l
f
r
a
ud a
s
non
-
f
r
a
ud
ul
e
nt
.
T
o
a
ddr
e
s
s
th
e
s
e
v
e
r
e
im
ba
la
nc
e
be
twe
e
n
f
r
a
udul
e
nt
a
nd
non
-
f
r
a
udul
e
nt
c
a
s
e
s
in
th
e
da
ta
s
e
t,
w
e
e
m
pl
oy
a
m
in
or
it
y
ove
r
s
a
m
pl
in
g
te
c
hni
que
c
a
ll
e
d
S
M
O
T
E
[
2
5]
.
T
hi
s
te
c
hni
qu
e
is
n
e
c
e
s
s
a
r
y
to
a
ddr
e
s
s
th
e
c
ha
ll
e
nge
w
h
e
r
e
th
e
m
in
or
it
y
c
la
s
s
is
of
te
n
ne
gl
e
c
te
d;
f
or
e
xa
m
pl
e
,
f
r
a
udul
e
nt
c
a
s
e
s
r
e
pr
e
s
e
nt
onl
y
0.67%
of
th
e
popula
ti
on,
w
hi
c
h
is
th
e
f
oc
u
s
of
our
a
tt
e
nt
io
n.
S
M
O
T
E
ge
ne
r
a
te
s
ne
w
s
ynt
he
ti
c
da
t
a
th
r
ough
a
n
it
e
r
a
ti
ve
pr
oc
e
s
s
ta
r
ge
ti
ng
e
a
c
h
poi
nt
in
th
e
m
in
or
it
y
c
la
s
s
.
I
t
pr
ove
s
to
be
a
n
e
f
f
e
c
ti
ve
m
e
th
od
f
or
a
ddr
e
s
s
in
g
e
xi
s
ti
ng
im
ba
la
nc
e
c
a
s
e
s
a
nd i
m
pr
ovi
ng c
la
s
s
if
ic
a
ti
on pe
r
f
or
m
a
nc
e
[
26]
.
3.
2
.
P
r
op
os
e
d
ar
t
if
ic
ia
l
n
e
u
r
al
n
e
t
w
or
k
s
T
hi
s
s
tu
dy
e
xp
e
r
im
e
nt
e
d
w
it
h
th
e
ut
il
iz
a
ti
on
of
ANN
,
s
pe
c
if
ic
a
ll
y
f
e
e
df
or
w
a
r
d
ne
twor
ks
,
to
de
te
c
t
f
r
a
udul
e
nt
f
in
a
nc
ia
l
s
ta
te
m
e
nt
s
f
r
om
a
s
e
v
e
r
e
ly
i
m
ba
la
nc
e
d da
ta
s
e
t.
T
he
e
xpe
r
im
e
nt
s
i
nvol
ve
d
t
hr
e
e
m
ode
l
s
or
ne
twor
ks
,
e
a
c
h
r
e
pr
e
s
e
nt
in
g
di
f
f
e
r
e
nt
f
r
a
ud
pr
e
di
c
to
r
s
a
s
th
e
in
p
ut
la
ye
r
.
T
he
f
ir
s
t
ne
twor
k
us
e
d
28
r
a
w
f
in
a
nc
ia
l
da
ta
a
s
in
put
la
y
e
r
s
de
r
iv
e
d
f
r
om
th
e
f
r
a
ud
pr
e
di
c
to
r
s
of
C
e
c
c
hi
ni
e
t
al
.
[
14]
.
T
he
s
e
c
ond
ne
twor
k
us
e
d
14
f
in
a
nc
ia
l
r
a
ti
os
a
s
in
put
la
ye
r
s
de
r
iv
e
d
f
r
om
th
e
f
r
a
ud
pr
e
di
c
to
r
s
of
D
e
c
how
e
t
al
.
[
16]
.
L
a
s
tl
y,
th
e
th
ir
d
ne
two
r
k
us
e
d a
c
om
bi
ne
d a
ppr
oa
c
h f
r
om
C
e
c
c
hi
ni
a
nd D
e
c
how
a
s
t
he
i
n
put
l
a
ye
r
.
T
he
ove
r
a
ll
a
r
c
hi
te
c
tu
r
e
of
th
e
pr
opos
e
d
ne
twor
ks
i
s
il
lu
s
tr
a
te
d
in
F
ig
ur
e
2.
T
he
a
r
c
hi
te
c
tu
r
e
of
th
e
th
r
e
e
ne
twor
ks
c
om
pr
is
e
s
a
n
in
put
la
ye
r
f
ol
lo
w
e
d
by
th
r
e
e
hi
dd
e
n
la
ye
r
s
a
nd
a
n
out
put
la
ye
r
.
T
he
in
put
la
ye
r
e
nc
om
pa
s
s
e
s
th
r
e
e
di
f
f
e
r
e
nt
s
c
e
na
r
io
s
,
r
e
pr
e
s
e
nt
in
g
di
f
f
e
r
e
nt
f
r
a
ud
pr
e
di
c
to
r
s
,
w
hi
c
h
c
a
n
be
r
e
pr
e
s
e
nt
e
d
by
in
put
l
a
ye
r
s
of
28, 14, a
nd 42, r
e
s
pe
c
ti
ve
ly
.
I
ns
pi
r
e
d
in
[
23]
–
[
25]
, t
he
f
ir
s
t
a
nd s
e
c
ond hidde
n l
a
ye
r
s
c
ons
i
s
t
of
f
ul
ly
c
onne
c
te
d l
a
ye
r
s
w
it
h L
e
a
kyR
e
L
U
(
a
lp
ha
of
0.05)
a
s
t
he
a
c
ti
va
ti
on f
unc
ti
on t
o a
ddr
e
s
s
c
om
pl
e
x pa
tt
e
r
ns
a
nd r
e
la
ti
ons
hi
ps
of
t
he
f
r
a
ud pr
e
di
c
to
r
s
a
nd h
a
ndl
e
non
-
li
ne
a
r
it
y i
s
s
ue
s
.
T
hi
s
i
s
f
ol
lo
w
e
d by L
2 r
e
gul
a
r
iz
a
ti
on
w
it
h
a
c
oe
f
f
ic
ie
nt
of
0.005
to
a
dd
a
pe
n
a
lt
y
te
r
m
to
th
e
ne
two
r
k
to
a
voi
d
ove
r
f
it
ti
ng
is
s
ue
s
[
27]
.
T
he
A
d
a
m
opt
im
iz
e
r
is
c
hos
e
n
f
or
it
s
c
a
pa
bi
li
ti
e
s
of
e
f
f
ic
ie
nt
c
om
put
a
ti
on, mi
ni
m
a
l
m
e
m
or
y
r
e
qui
r
e
m
e
nt
s
,
a
nd
s
ui
ta
bi
li
ty
f
or
la
r
ge
da
ta
s
e
ts
[
28]
.
A
ddi
ti
ona
ll
y,
a
dr
opout
la
ye
r
is
a
dde
d w
it
h
a
r
a
te
of
0.7
to
r
a
ndoml
y
dr
op
out
n
e
ur
ons
in
a
n
a
tt
e
m
pt
to
pr
e
ve
nt
ove
r
f
it
ti
ng
[
29]
.
F
in
a
ll
y,
a
n
out
put
la
ye
r
w
it
h
a
s
ig
m
oi
d
f
unc
ti
on
is
a
dde
d
to
pe
r
f
or
m
bi
na
r
y c
la
s
s
if
ic
a
ti
on t
a
s
k
s
.
3.
3.
P
e
r
f
or
m
an
c
e
e
val
u
at
io
n
T
he
pe
r
f
or
m
a
nc
e
of
th
e
pr
o
po
s
e
d
m
o
de
l
i
s
e
v
a
lu
a
te
d
us
in
g
t
hr
e
e
m
e
tr
ic
s
:
A
U
C
,
s
e
n
s
it
iv
i
ty
,
a
nd
pr
e
c
i
s
io
n.
A
U
C
i
s
a
m
e
tr
i
c
u
s
e
d
to
e
va
l
ua
te
th
e
pe
r
f
or
m
a
nc
e
of
bi
na
r
y
c
l
a
s
s
if
i
c
a
ti
on
[
30]
.
I
t
is
e
m
p
lo
ye
d
to
a
s
s
e
s
s
th
e
a
c
c
ur
a
c
y
of
t
he
pr
opo
s
e
d
m
ode
l
du
e
to
th
e
im
ba
la
n
c
e
i
n
t
he
d
a
ta
s
e
t,
w
h
e
r
e
t
he
oc
c
ur
r
e
n
c
e
of
f
r
a
ud
ul
e
nt
s
a
m
pl
e
s
is
not
a
de
qu
a
te
l
y
c
a
p
tu
r
e
d
in
s
ta
n
da
r
d
a
c
c
ur
a
c
y
m
e
tr
ic
s
[
31]
.
T
h
e
r
e
f
or
e
,
th
e
A
U
C
s
c
or
e
pr
ovi
de
s
a
m
or
e
r
e
pr
e
s
e
nt
a
ti
ve
m
e
a
s
ur
e
of
a
c
c
ur
a
c
y i
n
t
hi
s
c
o
nt
e
xt
c
om
pa
r
e
d
t
o
c
om
m
onl
y u
s
e
d a
c
c
ur
a
c
y
s
c
or
e
s
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
nt
J
A
r
ti
f
I
nt
e
ll
I
S
S
N
:
2252
-
8938
D
e
te
c
ti
ng f
r
audule
nt
f
in
anc
ia
l
s
ta
te
m
e
nt
und
e
r
i
m
bal
anc
e
d dat
a us
in
g ne
ur
al
ne
tw
o
r
k
(
Y
os
ua E
fr
ai
m
Y
oung
)
4109
F
ol
lo
w
in
g
pr
e
vi
ous
w
or
ks
[
15]
,
th
e
m
e
a
s
ur
e
m
e
nt
of
s
e
ns
it
iv
it
y
a
nd
pr
e
c
is
io
n
is
ba
s
e
d
on
da
ta
f
r
om
th
e
to
p
1%
of
obs
e
r
va
ti
ons
f
r
om
th
e
de
c
is
io
n
va
lu
e
.
T
hi
s
c
h
oi
c
e
is
dr
iv
e
n
by
pr
a
c
ti
c
a
l
c
on
s
id
e
r
a
ti
ons
,
a
s
r
e
gul
a
to
r
s
m
a
y
not
be
a
bl
e
to
obs
e
r
ve
a
ll
c
om
pa
ni
e
s
pr
e
di
c
t
e
d
a
s
f
r
a
udul
e
nt
due
to
r
e
s
our
c
e
c
on
s
tr
a
in
ts
.
A
ddi
ti
ona
ll
y,
th
is
de
c
is
io
n
is
in
f
lu
e
nc
e
d
by
th
e
r
e
s
ul
ts
of
le
a
di
ng
r
e
s
e
a
r
c
h
by
C
e
c
c
hi
ni
e
t
al
.
[
14]
,
w
hi
c
h
r
e
por
te
d
a
hi
gh
num
be
r
of
f
a
ls
e
pos
it
iv
e
s
in
th
e
ir
S
V
M
p
e
r
f
or
m
a
nc
e
,
c
or
r
e
c
tl
y
c
la
s
s
if
yi
ng
80%
o
f
f
r
a
ud
c
a
s
e
s
a
nd
90.6%
of
non
-
f
r
a
ud
c
a
s
e
s
.
T
he
r
e
f
or
e
,
to
m
it
ig
a
te
th
e
a
ll
oc
a
ti
on
of
e
xc
e
s
s
iv
e
r
e
s
our
c
e
s
to
w
a
r
d
m
a
ny
f
a
ls
e
pos
it
iv
e
s
, t
he
f
oc
us
i
s
on t
he
t
op 1%
of
obs
e
r
va
ti
ons
.
F
ig
ur
e
2
.
T
he
a
r
c
hi
te
c
tu
r
e
of
t
he
pr
opos
e
d ne
twor
ks
4.
R
E
S
U
L
T
S
A
N
D
D
I
S
C
U
S
S
I
O
N
T
hi
s
s
tu
dy
pr
opos
e
d
a
n
a
ppr
oa
c
h
f
or
de
te
c
ti
ng
f
r
a
udul
e
nt
f
in
a
nc
ia
l
s
ta
te
m
e
nt
s
by
c
om
bi
ni
ng
ne
ur
a
l
ne
twor
ks
a
nd
S
M
O
T
E
.
W
hi
le
e
a
r
li
e
r
li
te
r
a
tu
r
e
ha
s
e
xpl
or
e
d
a
nd
de
m
ons
tr
a
te
d
th
e
c
a
pa
bi
li
ty
of
va
r
io
us
a
lg
or
it
hm
s
to
de
te
c
t
f
r
a
udul
e
nt
f
in
a
nc
ia
l
s
ta
te
m
e
nt
s
,
th
e
r
e
is
s
ti
l
l
e
nha
nc
e
m
e
nt
ne
e
de
d
s
p
e
c
if
ic
a
ll
y
in
te
r
m
s
of
a
c
c
ur
a
c
y.
T
he
r
e
f
or
e
,
de
ve
lo
pi
ng
a
m
ode
l
w
it
h
be
tt
e
r
a
c
c
ur
a
c
y
is
im
por
ta
nt
,
e
s
pe
c
ia
ll
y
a
m
ode
l
th
a
t
c
a
n
be
r
e
li
e
d upon f
or
pr
a
c
ti
c
a
l
a
dopt
io
n.
T
hr
e
e
ne
twor
ks
w
it
h
di
f
f
e
r
e
nt
f
r
a
ud
pr
e
di
c
to
r
a
s
in
put
la
ye
r
a
r
e
e
m
pl
oye
d.
T
he
r
e
s
ul
ts
obt
a
in
e
d
a
r
e
s
um
m
a
r
iz
e
d
in
T
a
bl
e
1.
I
n
th
e
f
ir
s
t
ne
twor
k
,
e
m
pl
oyi
ng
r
a
w
f
in
a
nc
ia
l
da
ta
a
s
f
r
a
ud
pr
e
di
c
to
r
,
th
e
pr
opo
s
e
d
ne
twor
k
s
c
or
e
d
a
n
A
U
C
s
c
or
e
of
0.706,
w
it
h
a
s
e
n
s
it
iv
it
y
of
50%
a
nd
1.39%
.
T
he
s
e
c
ond
pr
opo
s
e
d
ne
twor
k
a
c
hi
e
ve
d
A
U
C
s
c
or
e
of
0.693,
s
e
n
s
it
iv
it
y
of
67%
,
a
nd
pr
e
c
is
io
n
of
2.89%
,
by
e
m
pl
oyi
ng
f
in
a
nc
ia
l
r
a
ti
os
a
s
f
r
a
ud
pr
e
di
c
to
r
.
L
a
s
tl
y,
th
e
th
ir
d
ne
twor
k
e
m
pl
oyi
ng
bot
h
r
a
w
f
in
a
nc
ia
l
da
ta
a
nd
f
in
a
nc
ia
l
r
a
ti
o
s
a
s
f
r
a
ud
pr
e
di
c
to
r
r
e
s
ul
te
d i
n A
U
C
of
0.672, f
ol
lo
w
e
d by s
e
ns
it
iv
it
y of
83%
a
nd pr
e
c
is
io
n of
1.91%
.
T
a
bl
e
1
. S
um
m
a
r
y of
c
om
pa
r
is
on (
t
e
s
t
pe
r
io
d 2003
–
2008)
F
r
a
ud pr
e
di
c
t
or
AUC
S
e
ns
i
t
i
vi
t
y
(%)
P
r
e
c
i
s
i
on
(%)
28 r
a
w
f
i
na
nc
i
a
l
da
t
a
0.706
50
1.39
14 f
i
na
nc
i
a
l
r
a
t
i
os
0.693
67
2.89
B
ot
h
0.672
83
1.91
A
s
pr
e
s
e
nt
e
d
in
T
a
bl
e
1,
it
is
f
ound
th
a
t
th
e
pr
opos
e
d
ne
ur
a
l
ne
twor
k
c
a
n
de
te
c
t
f
r
a
udul
e
nt
f
in
a
nc
ia
l
s
ta
te
m
e
nt
s
,
w
it
h
r
a
w
f
in
a
nc
ia
l
da
ta
i
s
th
e
be
s
t
f
r
a
ud
pr
e
di
c
to
r
m
e
a
s
ur
e
d
by
A
U
C
s
c
or
e
.
T
he
s
e
r
e
s
ul
ts
de
m
ons
tr
a
te
d
th
a
t
c
om
bi
ni
ng both r
a
w
f
in
a
nc
ia
l
da
ta
a
nd f
in
a
nc
ia
l
r
a
ti
os
a
s
f
r
a
ud pr
e
di
c
to
r
doe
s
not
yi
e
ld
t
o a
hi
ghe
r
a
c
c
ur
a
c
y,
m
e
a
s
ur
e
d
by
A
U
C
s
c
or
e
.
F
r
om
th
is
e
xpe
r
im
e
nt
a
ti
on,
t
he
hi
ghe
s
t
pr
e
c
is
io
n
is
obt
a
in
e
d
th
r
ough
e
m
pl
oyi
ng
f
in
a
nc
ia
l
r
a
ti
os
a
s
f
r
a
ud
pr
e
di
c
to
r
,
w
he
r
e
a
s
th
e
hi
g
he
s
t
s
e
n
s
it
iv
it
y
is
obt
a
in
e
d
th
r
ough
e
m
pl
oyi
ng
bot
h r
a
w
f
in
a
nc
ia
l
da
ta
a
nd f
in
a
nc
ia
l
r
a
ti
os
a
s
f
r
a
ud pr
e
di
c
to
r
.
T
a
bl
e
2
pr
ovi
de
s
r
e
s
ul
t
s
obt
a
in
e
d
f
r
om
th
e
pr
opo
s
e
d
ne
twor
k
f
or
th
e
te
s
t
pe
r
io
d
of
2003
–
2008
in
c
om
pa
r
is
on
w
it
h
pr
e
vi
ous
li
te
r
a
tu
r
e
.
T
he
s
e
r
e
s
ul
ts
d
e
m
ons
tr
a
te
t
ha
t
th
e
pr
opos
e
d
ne
twor
k
h
a
s
c
om
p
a
r
a
bl
e
A
U
C
s
c
or
e
a
nd
pr
e
c
i
s
io
n
in
c
om
pa
r
is
on
w
it
h
r
e
s
ul
ts
obt
a
in
e
d
f
r
om
pr
e
vi
ous
li
te
r
a
tu
r
e
s
by
e
m
pl
oyi
ng
a
lg
or
it
hm
s
w
it
h
S
V
M
,
LR
,
R
U
S
B
oos
t,
X
G
B
oos
t,
a
nd
f
-
X
G
B
oos
t.
I
n
c
o
nt
r
a
s
t,
th
e
pr
opos
e
d
ne
twor
k
de
m
on
s
tr
a
te
s
a
s
upe
r
io
r
s
e
ns
it
iv
it
y s
c
or
e
,
in
di
c
a
ti
ng
th
a
t
th
e
m
od
e
l
is
a
bl
e
to
id
e
nt
if
y
f
r
a
ud
w
it
hout
pr
oduc
in
g
a
hi
gh
num
be
r
of
f
a
ls
e
ne
ga
ti
ve
s
, w
hi
c
h
c
oul
d
tr
a
ns
la
te
to
unde
te
c
te
d
f
r
a
ud.
H
e
nc
e
,
th
is
de
m
on
s
tr
a
te
s
th
e
m
od
e
l’
s
a
s
s
ur
a
nc
e
by
e
ns
ur
in
g
r
e
li
a
bi
li
ty
a
nd
pr
a
c
ti
c
a
l
ut
i
li
ty
o
r
a
dopt
io
n.
T
o
s
how
r
obus
tn
e
s
s
of
th
e
pr
opos
e
d
ne
twor
k,
th
r
e
e
a
lt
e
r
na
ti
ve
te
s
t
pe
r
io
ds
a
r
e
a
dde
d.
C
on
s
is
te
nt
w
it
h
pr
e
vi
ous
s
t
udy
[
15]
,
[
22
]
,
th
e
a
dd
it
io
na
l
te
s
t
pe
r
io
ds
a
r
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
2252
-
8938
I
nt
J
A
r
ti
f
I
nt
e
ll
,
V
ol
.
13
, N
o.
4
,
D
e
c
e
m
be
r
2024
:
4106
-
4112
4110
2003
–
2005, 2003
–
2011, a
nd
2003
–
2014.
T
he
nume
r
ic
a
l
f
ig
ur
e
of
pe
r
f
or
m
a
nc
e
m
e
tr
ic
s
f
or
di
f
f
e
r
e
nt
a
ddi
ti
ona
l
te
s
t
pe
r
io
ds
a
r
e
pr
e
s
e
nt
e
d i
n T
a
bl
e
s
3
to
5.
T
a
bl
e
2
.
S
um
m
a
r
y of
c
om
pa
r
is
on w
it
h pr
e
vi
ous
s
tu
dy (
t
e
s
t
pe
r
i
od 2003
–
2008)
F
r
a
ud pr
e
di
c
t
or
M
ode
l
AUC
S
e
ns
i
t
i
vi
t
y
(%)
P
r
e
c
i
s
i
on
(%)
R
e
f
e
r
e
nc
e
28 r
a
w
f
i
na
nc
i
a
l
da
t
a
S
V
M
0.626
2.53
1.92
[
15]
L
ogi
s
t
i
c
0.690
0.73
0.85
[
15]
R
U
S
B
oos
t
0.725
4.88
4.48
[
15]
X
G
B
oos
t
0.689
3.56
3.36
[
22]
f
-
X
G
B
oos
t
0.693
5.00
4.22
[
22]
ANN
0.706
50
1.39
T
hi
s
s
t
udy
14
f
i
na
nc
i
a
l
r
a
t
i
os
L
ogi
s
t
i
c
0.672
3.99
2.63
[
15]
ANN
0.693
67
2.89
T
hi
s
s
t
udy
B
ot
h
R
U
S
B
oos
t
0.696
3.19
2.54
[
15]
ANN
0.672
83
1.91
T
hi
s
s
t
udy
T
a
bl
e
3
.
S
um
m
a
r
y of
c
om
pa
r
is
on w
it
h pr
e
vi
ous
s
tu
dy (
t
e
s
t
pe
r
i
o
d 2003
–
2005)
F
r
a
ud pr
e
di
c
t
or
M
ode
l
AUC
S
e
ns
i
t
i
vi
t
y
(%)
P
r
e
c
i
s
i
on
(%)
R
e
f
e
r
e
nc
e
28 r
a
w
f
i
na
nc
i
a
l
da
t
a
S
V
M
0.637
2.28
2.53
[
15]
L
ogi
s
t
i
c
0.685
1.45
1.69
[
15]
R
U
S
B
oos
t
0.753
7.64
7.83
[
15]
f
-
X
G
B
oos
t
0.691
6.59
6.71
[
22]
ANN
0.694
100
2.79
T
hi
s
s
t
udy
14 f
i
na
nc
i
a
l
r
a
t
i
os
L
ogi
s
t
i
c
0.649
1.37
1.29
[
15]
ANN
0.667
67
2.53
T
hi
s
s
t
udy
B
ot
h
ANN
0.656
100
2.52
T
hi
s
s
t
udy
T
a
bl
e
4
.
S
um
m
a
r
y of
c
om
pa
r
is
on w
it
h pr
e
vi
ous
s
tu
dy (
t
e
s
t
pe
r
i
od 2003
–
2
0
11
)
F
r
a
ud pr
e
di
c
t
or
M
ode
l
AUC
S
e
ns
i
t
i
vi
t
y
(%)
P
r
e
c
i
s
i
on
(%)
R
e
f
e
r
e
nc
e
28 r
a
w
f
i
na
nc
i
a
l
da
t
a
S
V
M
0.647
3.07
1.98
[
15]
L
ogi
s
t
i
c
0.702
1.87
1.19
[
15]
R
U
S
B
oos
t
0.710
4.40
3.60
[
15]
f
-
X
G
B
oos
t
0.678
3.69
3.02
[
22]
ANN
0.720
56
1.34
T
hi
s
s
t
udy
14 f
i
na
nc
i
a
l
r
a
t
i
os
L
ogi
s
t
i
c
0.672
3.49
2.23
[
15]
ANN
0.685
67
2.40
T
hi
s
s
t
udy
B
ot
h
ANN
0.693
89
2.45
T
hi
s
s
t
udy
T
a
bl
e
5
.
S
um
m
a
r
y of
c
om
pa
r
is
on w
it
h pr
e
vi
ous
s
tu
dy (
t
e
s
t
pe
r
i
od 2003
–
20
14
)
F
r
a
ud pr
e
di
c
t
or
M
ode
l
AUC
S
e
ns
i
t
i
vi
t
y
(%)
P
r
e
c
i
s
i
on
(%)
R
e
f
e
r
e
nc
e
28 r
a
w
f
i
na
nc
i
a
l
da
t
a
S
V
M
0.628
2.30
1.48
[
15]
L
ogi
s
t
i
c
0.709
1.84
1.04
[
15]
R
U
S
B
oos
t
0.717
3.30
2.70
[
15]
f
-
X
G
B
oos
t
0.678
2.77
2.26
[
22]
ANN
0.718
50
1.15
T
hi
s
s
t
udy
14 f
i
na
nc
i
a
l
r
a
t
i
os
L
ogi
s
t
i
c
0.702
3.45
1.86
[
15]
ANN
0.694
58
1.99
T
hi
s
s
t
udy
B
ot
h
ANN
0.686
75
2.02
T
hi
s
s
t
udy
T
he
r
e
s
ul
ts
obt
a
in
e
d a
r
e
c
om
pa
r
e
d w
it
h pr
e
vi
ous
l
it
e
r
a
tu
r
e
a
nd
s
um
m
a
r
iz
e
d i
n F
ig
ur
e
3. A
s
s
how
n i
n
F
ig
ur
e
3,
S
V
M
m
ode
l
de
m
on
s
tr
a
te
f
lu
c
tu
a
ti
ons
in
th
e
pe
r
f
o
r
m
a
nc
e
w
hi
le
bot
h
R
U
S
B
oos
t
a
nd
X
G
B
oos
t
de
m
ons
tr
a
te
a
p
e
r
f
or
m
a
nc
e
de
c
li
ne
w
h
e
n
th
e
r
a
nge
of
th
e
s
e
t
is
e
xt
e
nde
d.
T
he
r
e
s
ul
ts
i
s
a
c
c
or
d
to
H
oa
ng
e
t
al
.
[
22
]
,
th
a
t
th
e
a
s
s
um
pt
io
n
of
unde
te
c
te
d
f
r
a
ud
gr
o
w
s
ove
r
ti
m
e
m
a
ke
s
a
lo
nge
r
te
s
t
pe
r
io
d
le
s
s
r
e
li
a
bl
e
.
H
ow
e
ve
r
,
in
c
ont
r
a
s
t
to
S
V
M
,
R
U
S
B
oos
t
,
a
nd
X
G
B
o
os
t
m
ode
l,
th
e
pr
opos
e
d
ne
twor
k
,
a
nd
l
og
is
ti
c
m
ode
ls
s
how
a
s
li
ght
ly
pe
r
f
or
m
a
nc
e
im
pr
ove
m
e
nt
f
or
th
e
e
xt
e
ns
io
n
of
th
e
te
s
t
pe
r
io
d.
T
hi
s
d
e
m
ons
tr
a
te
s
th
e
r
obus
tn
e
s
s
of
bot
h m
ode
ls
a
nd i
s
e
xpe
c
t
e
d t
o ha
ve
a
s
ta
bl
e
pe
r
f
or
m
a
nc
e
w
he
n t
e
s
te
d w
it
h ne
w
uns
e
e
n da
ta
.
E
m
pl
oyi
ng
r
a
w
f
in
a
nc
ia
l
da
ta
a
s
f
r
a
ud
pr
e
di
c
to
r
,
th
e
pr
opos
e
d
ne
twor
k
de
m
ons
tr
a
te
d
th
e
be
s
t
A
U
C
s
c
or
e
in
s
c
e
na
r
io
o
f
us
in
g
f
ul
l
te
s
t
s
e
t
of
2003
–
2014
a
s
s
how
n
in
T
a
bl
e
5
by
s
c
or
in
g
A
U
C
of
0.718
w
it
h
pr
e
c
is
io
n
of
1.15%
,
a
nd
s
e
ns
it
iv
it
y
o
f
50%
.
C
ons
id
e
r
in
g
s
ta
bi
li
ty
of
A
U
C
to
de
m
ons
tr
a
te
r
obus
tn
e
s
s
,
th
e
pr
opos
e
d
ne
twor
k
s
c
or
e
A
U
C
of
0.694
a
nd
0.720,
in
te
s
t
pe
r
io
d
of
2003
–
2005
a
nd
2003
–
2011
a
s
s
how
n
in
T
a
bl
e
s
3
a
nd
4,
r
e
s
pe
c
ti
ve
ly
.
T
hi
s
s
how
s
th
a
t
e
xp
a
ndi
ng
da
t
a
s
e
t
im
pr
ove
s
th
e
pe
r
f
or
m
a
nc
e
of
th
e
pr
opos
e
d
ne
t
w
or
k
.
T
he
n,
in
th
e
ne
xt
s
c
e
na
r
io
us
in
g
pe
r
io
d
of
2003
–
2014,
th
e
A
U
C
s
c
or
e
dr
oppe
d
to
0.718,
s
li
ght
ly
lo
w
e
r
t
ha
n
pr
e
vi
ous
s
c
e
na
r
io
of
2003
–
2011.
T
hi
s
in
di
c
a
te
s
th
a
t
w
hi
le
e
xpa
ndi
ng
da
ta
s
e
t
im
pr
ove
s
th
e
pe
r
f
or
m
a
nc
e
,
Evaluation Warning : The document was created with Spire.PDF for Python.
I
nt
J
A
r
ti
f
I
nt
e
ll
I
S
S
N
:
2252
-
8938
D
e
te
c
ti
ng f
r
audule
nt
f
in
anc
ia
l
s
ta
te
m
e
nt
und
e
r
i
m
bal
anc
e
d dat
a us
in
g ne
ur
al
ne
tw
o
r
k
(
Y
os
ua E
fr
ai
m
Y
oung
)
4111
th
e
r
e
m
a
y
be
a
di
m
in
is
hi
ng
r
e
tu
r
n
in
a
c
e
r
ta
in
le
ngt
h
of
p
e
r
io
ds
.
C
ons
i
s
te
nt
w
it
h
th
e
r
e
s
ul
t
s
of
B
a
o
e
t
al
.
[
15]
,
th
is
s
tu
dy
r
e
s
ul
t
s
de
m
ons
tr
a
te
d
th
a
t
w
he
n
e
xpe
r
im
e
nt
in
g
w
it
h
th
e
s
a
m
e
m
ode
l
or
ne
twor
ks
,
u
s
in
g
28
r
a
w
f
in
a
nc
ia
l
da
ta
th
a
t
d
e
r
iv
e
d
f
r
om
[
14]
le
a
ds
to
a
be
tt
e
r
r
e
s
ul
t
c
om
pa
r
e
d
to
us
in
g
th
e
ot
he
r
f
r
a
ud
pr
e
di
c
to
r
s
,
w
hi
c
h
is
14 f
in
a
nc
ia
l
r
a
ti
os
de
r
iv
e
d f
r
om
[
16]
.
T
hi
s
s
tu
dy
r
e
s
ul
ts
s
how
s
th
a
t
c
om
bi
ni
ng
ne
ur
a
l
ne
twor
k
a
nd
S
M
O
T
E
c
a
n
de
te
c
t
f
r
a
udul
e
nt
f
in
a
nc
ia
l
s
ta
te
m
e
nt
s
i
n a
s
e
ve
r
e
ly
i
m
ba
la
nc
e
d da
t
a
s
e
t
us
in
g r
a
w
f
in
a
nc
ia
l
da
ta
, f
in
a
nc
ia
l
r
a
ti
os
, or
bot
h
c
om
bi
ne
d a
s
t
he
f
r
a
ud
pr
e
di
c
to
r
.
W
hi
le
th
e
pr
opos
e
d
ne
twor
k
de
m
ons
tr
a
te
d
pr
om
is
in
g
ut
il
it
y,
it
is
im
por
ta
nt
to
a
c
kno
w
le
dge
th
a
t
th
e
da
ta
s
e
t
u
s
e
d c
on
s
is
ts
of
hi
s
to
r
ic
a
l
da
t
a
t
ha
t
c
om
in
g f
r
om
s
pe
c
if
ic
d
e
m
ogr
a
phi
c
s
a
nd
t
im
e
pe
r
io
ds
.
T
hi
s
m
a
y
pr
om
ot
e
li
m
it
a
ti
ons
on
ge
ne
r
a
li
z
a
bi
li
ty
,
he
nc
e
r
e
qui
r
e
f
ur
th
e
r
c
a
li
br
a
ti
on
or
up
da
te
s
to
m
a
in
ta
in
it
s
e
f
f
e
c
ti
ve
ne
s
s
i
n t
he
c
ur
r
e
nt
dyna
m
ic
e
nvi
r
onm
e
nt
.
F
ig
ur
e
3
.
S
um
m
a
r
y of
A
U
C
s
c
or
e
s
ove
r
t
he
a
ddi
ti
ona
l
te
s
t
pe
r
i
ods
5.
C
O
N
C
L
U
S
I
O
N
T
hi
s
pa
pe
r
in
tr
oduc
e
s
a
ne
ur
a
l
ne
twor
k
de
s
ig
ne
d
to
de
t
e
c
t
f
r
a
udul
e
nt
f
in
a
nc
ia
l
s
ta
te
m
e
nt
s
w
it
hi
n
a
n
im
ba
la
nc
e
d da
ta
s
e
t,
a
ddr
e
s
s
in
g t
he
s
e
ve
r
e
i
m
ba
la
nc
e
i
s
s
ue
t
hr
ough the
ut
il
iz
a
ti
on of
S
M
O
T
E
. O
ur
e
xpe
r
im
e
nt
r
e
s
ul
ts
in
di
c
a
te
th
a
t
th
e
m
od
e
l
a
c
hi
e
ve
s
de
te
c
ti
o
n
c
a
pa
bi
li
ti
e
s
, w
it
h
a
n
A
U
C
s
c
or
e
of
70.6%
,
a
s
e
n
s
it
iv
it
y
r
a
te
of
83%
,
a
nd
a
pr
e
c
is
io
n
r
a
te
of
2.89%
.
T
hi
s
s
tu
dy
c
ont
r
ib
ut
e
s
s
i
gni
f
ic
a
nt
ly
by
a
dvoc
a
ti
ng
f
or
th
e
in
te
gr
a
ti
on
of
ANN
in
a
udi
ti
ng pr
a
c
ti
c
e
s
, pa
r
ti
c
ul
a
r
ly
dur
in
g t
he
i
ni
ti
a
l
a
udi
t
p
ha
s
e
, s
uc
h
a
s
r
is
k a
s
s
e
s
s
m
e
nt
pr
oc
e
dur
e
s
.
T
he
pr
opos
e
d
m
ode
l'
s
hi
gh
s
e
ns
it
iv
it
y
r
a
te
unde
r
s
c
or
e
s
it
s
s
upe
r
io
r
it
y
ove
r
s
im
il
a
r
m
ode
ls
,
of
f
e
r
in
g
pr
a
c
ti
c
a
l
ut
il
i
ty
f
or
a
udi
to
r
s
a
nd r
e
gul
a
to
r
s
by mi
ni
m
iz
in
g t
he
r
is
k of
f
a
ls
e
ne
ga
t
iv
e
s
. H
ow
e
ve
r
, l
im
it
a
ti
ons
e
xi
s
t,
i
nc
lu
di
ng
th
e
r
e
li
a
nc
e
s
ol
e
ly
on nume
r
ic
a
l
f
in
a
nc
ia
l
da
ta
e
xt
r
a
c
t
e
d f
r
om
f
in
a
nc
ia
l
s
ta
te
m
e
nt
s
. F
ut
ur
e
r
e
s
e
a
r
c
h a
ve
nue
s
c
oul
d
e
xpl
or
e
th
e
c
om
bi
na
ti
on
of
non
-
f
in
a
nc
ia
l
da
ta
a
nd
th
e
a
ppl
ic
a
ti
on
of
uns
upe
r
vi
s
e
d
le
a
r
ni
ng
to
a
ddr
e
s
s
m
is
la
be
li
ng
is
s
ue
s
,
pot
e
nt
ia
ll
y
th
r
ough
th
e
im
pl
e
m
e
nt
a
ti
on
of
g
e
ne
r
a
ti
ve
a
r
ti
f
ic
ia
l
in
te
ll
ig
e
nc
e
to
ge
ne
r
a
te
f
r
a
udul
e
nt
da
ta
f
or
t
r
a
in
in
g pur
pos
e
s
or
de
s
c
r
ib
in
g f
r
a
ud c
ha
r
a
c
te
r
is
ti
c
s
.
R
E
F
E
R
E
N
C
E
S
[
1]
“
I
n
t
e
r
na
t
i
ona
l
s
t
a
n
da
r
d
on
a
ud
i
t
i
n
g
240
:
t
he
a
u
di
t
o
r
’
s
r
e
s
po
ns
i
bi
l
i
t
i
e
s
r
e
l
a
t
i
ng
t
o
f
r
a
ud
i
n
a
n
a
ud
i
t
o
f
f
i
na
nc
i
a
l
s
t
a
t
e
m
e
nt
s
,
”
I
F
A
C
,
20
13.
A
c
c
e
s
s
e
d
:
D
e
c
.
2
7,
2
023
.
[
O
nl
i
ne
]
.
A
va
i
l
a
bl
e
:
ht
t
ps
:
/
/
w
w
w
.i
f
a
c
.o
r
g/
_
f
l
ys
ys
t
e
m
/
a
z
u
r
e
-
p
r
i
va
t
e
/
pu
bl
i
c
a
t
i
o
ns
/
f
i
l
e
s
/
A
0
12
2
013
I
A
A
S
B
H
a
nd
boo
k
I
S
A
24
0.p
df
.
[
2]
Y
.
J
.
C
he
n,
W
.
C
.
L
i
ou,
Y
.
M
.
C
he
n,
a
nd
J
.
H
.
W
u,
“
F
r
a
ud
de
t
e
c
t
i
on
f
or
f
i
na
nc
i
a
l
s
t
a
t
e
m
e
nt
s
of
bus
i
n
e
s
s
gr
oup
s
,”
I
nt
e
r
nat
i
onal
J
our
nal
of
A
c
c
ount
i
ng I
nf
or
m
at
i
on Sy
s
t
e
m
s
,
vol
. 32, pp. 1
–
23, 2019
,
doi
:
10.10
16/
j
.a
c
c
i
nf
.2018.11.004
.
[
3]
M
.
N
.
A
s
ht
i
a
ni
a
nd
B
.
R
a
a
he
m
i
,
“
I
nt
e
l
l
i
ge
nt
f
r
a
ud
de
t
e
c
t
i
on
i
n
f
i
na
nc
i
a
l
s
t
a
t
e
m
e
nt
s
us
i
ng
m
a
c
hi
ne
l
e
a
r
ni
ng
a
nd
da
t
a
m
i
ni
ng:
a
s
ys
t
e
m
a
t
i
c
l
i
t
e
r
a
t
ur
e
r
e
vi
e
w
,”
I
E
E
E
A
c
c
e
s
s
, vol
. 10, pp. 72504
–
72525, 2022, doi
:
10.1109/
A
C
C
E
S
S
.2021.3096799.
[
4]
W
.
X
i
uguo
a
nd
D
.
S
he
ngyong,
“
A
n
a
n
a
l
ys
i
s
on
f
i
na
nc
i
a
l
s
t
a
t
e
m
e
nt
f
r
a
ud
de
t
e
c
t
i
on
f
or
C
hi
ne
s
e
l
i
s
t
e
d
c
om
pa
ni
e
s
us
i
ng
de
e
p
l
e
a
r
ni
ng,”
I
E
E
E
A
c
c
e
s
s
, vol
. 10, pp. 22516
–
22532, 2022, doi
:
10.1109/
A
C
C
E
S
S
.2022.3153478.
[
5]
D
.
B
ot
e
z
,
“
R
e
c
e
nt
c
ha
l
l
e
nge
f
or
a
udi
t
or
s
:
us
i
ng
da
t
a
a
na
l
yt
i
c
s
i
n
t
he
a
udi
t
of
t
he
f
i
na
nc
i
a
l
s
t
a
t
e
m
e
nt
s
,”
B
r
ai
n
-
br
oad
R
e
s
e
ar
c
h
i
n
A
r
t
i
f
i
c
i
al
I
nt
e
l
l
i
ge
nc
e
and N
e
ur
os
c
i
e
n
c
e
, vol
. 9, no. 4, pp. 61
–
72, 2018
.
[
6]
G
.
S
a
l
i
j
e
ni
,
A
.
S
.
-
T
a
dde
i
,
a
nd
S
.
T
ur
l
e
y,
“
B
i
g
da
t
a
a
nd
c
ha
nge
s
i
n
a
udi
t
t
e
c
hnol
ogy:
c
ont
e
m
pl
a
t
i
ng
a
r
e
s
e
a
r
c
h a
ge
nda
,”
A
c
c
ount
i
ng
and B
us
i
ne
s
s
R
e
s
e
a
r
c
h
, vol
. 49, no. 1, pp. 95
–
119, 2019, doi
:
10.1080/
0001478
8.2018.1459458.
[
7]
B
. P
. G
r
e
e
n a
nd J
. H
. C
hoi
, “
A
s
s
e
s
s
i
ng t
he
r
i
s
k of
m
a
na
ge
m
e
nt
f
r
a
ud t
hr
ough n
e
ur
a
l
ne
t
w
or
k t
e
c
hnol
ogy,”
A
udi
t
i
ng
, vol
.
16, no. 1,
pp. 25
–
28, 1997.
[
8]
S
.
K
ot
s
i
a
nt
i
s
,
E
.
K
oum
a
na
kos
,
D
.
T
z
e
l
e
pi
s
,
a
nd
V
.
T
a
m
pa
ka
s
,
“
F
or
e
c
a
s
t
i
ng
f
r
a
udul
e
nt
f
i
na
nc
i
a
l
s
t
a
t
e
m
e
nt
s
us
i
ng
da
t
a
m
i
ni
ng,”
I
nt
e
r
nat
i
onal
J
our
nal
of
C
om
put
at
i
onal
I
nt
e
l
l
i
ge
nc
e
, vol
. 3, no. 2, pp. 104
–
110, 2006
.
[
9]
E
. K
i
r
kos
, C
. S
pa
t
hi
s
, a
nd
Y
. M
a
nol
opoul
o
s
, “
D
a
t
a
m
i
ni
ng t
e
c
hni
que
s
f
or
t
he
de
t
e
c
t
i
on of
f
r
a
udul
e
nt
f
i
na
nc
i
a
l
s
t
a
t
e
m
e
nt
s
,”
E
x
pe
r
t
Sy
s
t
e
m
s
w
i
t
h A
ppl
i
c
at
i
ons
, vol
. 32, no. 4, pp. 995
–
1003, 2007, doi
:
10.1016/
j
.e
s
w
a
.2006.02.016.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
2252
-
8938
I
nt
J
A
r
ti
f
I
nt
e
ll
,
V
ol
.
13
, N
o.
4
,
D
e
c
e
m
be
r
2024
:
4106
-
4112
4112
[
10]
J
. P
e
r
ol
s
, “
F
i
na
nc
i
a
l
s
t
a
t
e
m
e
nt
f
r
a
ud de
t
e
c
t
i
on:
a
n a
na
l
ys
i
s
of
s
t
a
t
i
s
t
i
c
a
l
a
nd m
a
c
hi
ne
l
e
a
r
ni
ng a
l
gor
i
t
hm
s
,”
A
udi
t
i
ng
, vol
. 30, no. 2,
pp. 19
–
50, 2011, doi
:
10.2308/
a
j
pt
-
50009.
[
11]
C
.
L
.
J
a
n,
“
D
e
t
e
c
t
i
on
of
f
i
na
nc
i
a
l
s
t
a
t
e
m
e
nt
f
r
a
ud
us
i
ng
de
e
p
l
e
a
r
ni
ng
f
or
s
us
t
a
i
na
bl
e
de
ve
l
opm
e
nt
of
c
a
pi
t
a
l
m
a
r
ke
t
s
unde
r
i
nf
or
m
a
t
i
on a
s
ym
m
e
t
r
y,”
Sus
t
ai
nabi
l
i
t
y
, vol
. 13, no. 17, pp. 9879
–
9898, 2021,
doi
:
10.3390/
s
u13179879.
[
12]
T
.
K
i
e
hl
,
B
.
H
oog
s
,
L
.
C
hr
i
s
t
i
na
,
a
nd
S
.
D
e
ni
z
,
“
E
vol
vi
ng
m
ul
t
i
-
va
r
i
a
t
e
t
i
m
e
-
s
e
r
i
e
s
p
a
t
t
e
r
ns
f
or
t
he
di
s
c
r
i
m
i
na
t
i
on
of
f
r
a
udul
e
nt
f
i
na
nc
i
a
l
f
i
l
i
ngs
,”
G
e
ne
t
i
c
and E
v
ol
ut
i
onar
y
C
om
put
at
i
on C
onf
e
r
e
nc
e
,
pp. 1
-
8,
2005.
[
13]
J
. B
e
r
t
om
e
u, E
. C
he
yne
l
,
E
. F
l
oyd, a
nd
W
. P
a
n,
“
U
s
i
ng m
a
c
hi
ne
l
e
a
r
ni
ng t
o
de
t
e
c
t
m
i
s
s
t
a
t
e
m
e
nt
s
,”
R
e
v
i
e
w
of
A
c
c
ount
i
ng St
udi
e
s
,
vol
. 26, no. 2, pp. 468
–
519, 2021, doi
:
10.1007/
s
11142
-
020
-
09563
-
8.
[
14]
M
.
C
e
c
c
hi
ni
,
H
.
A
yt
ug,
G
.
J
.
K
oe
hl
e
r
,
a
nd
P
.
P
a
t
ha
k,
“
D
e
t
e
c
t
i
ng
m
a
na
ge
m
e
nt
f
r
a
ud
i
n
publ
i
c
c
om
pa
ni
e
s
,
”
M
anage
m
e
nt
Sc
i
e
nc
e
,
vol
. 56, no. 7, pp. 1146
–
1160, 2010, doi
:
10.1287/
m
ns
c
.1100.1174.
[
15]
Y
.
B
a
o,
B
.
K
e
,
B
.
L
i
,
Y
.
J
.
Y
u,
a
nd
J
.
Z
ha
ng,
“
D
e
t
e
c
t
i
ng
a
c
c
ount
i
ng
f
r
a
ud
i
n
p
ubl
i
c
l
y
t
r
a
de
d
U
.S
.
f
i
r
m
s
us
i
ng
a
m
a
c
hi
ne
l
e
a
r
n
i
ng
a
ppr
oa
c
h,”
J
our
nal
of
A
c
c
ount
i
ng R
e
s
e
a
r
c
h
, vol
. 58, no. 1, pp. 199
–
235, 2020,
doi
:
10.1111/
1475
-
679X
.12292.
[
16]
P
. M
. D
e
c
how
,
W
. G
e
,
C
. R
.
L
a
r
s
on,
a
nd R
.
G
. S
l
oa
n,
“
P
r
e
di
c
t
i
ng m
a
t
e
r
i
a
l
a
c
c
ount
i
ng m
i
s
s
t
a
t
e
m
e
nt
s
,
”
39
th
A
nnual
C
ont
e
m
por
a
r
y
A
c
c
ount
i
ng R
e
s
e
ar
c
h
C
onf
e
r
e
nc
e
,
vol
. 28, no. 1,
pp.
17
-
82
,
2011
, doi
:
10.1111/
j
.1911
-
3846.2010.01041.x
.
[
17]
Q
.
D
e
ng
a
nd
G
.
M
e
i
,
“
C
om
bi
ni
ng
s
e
l
f
-
or
ga
ni
z
i
ng
m
a
p
a
nd
k
-
m
e
a
n
s
c
l
us
t
e
r
i
n
g
f
or
de
t
e
c
t
i
ng
f
r
a
udul
e
nt
f
i
na
nc
i
a
l
s
t
a
t
e
m
e
nt
s
,
”
i
n
2009 I
E
E
E
I
nt
e
r
nat
i
onal
C
onf
e
r
e
nc
e
on G
r
anul
a
r
C
om
put
i
ng, G
R
C
, 2009, pp.
126
–
131, doi
:
10.1109/
G
R
C
.2009.5255148.
[
18]
S
.
Y
.
H
ua
ng,
R
.
H
.
T
s
a
i
h,
a
nd
W
.
Y
.
L
i
n,
“
U
n
s
upe
r
vi
s
e
d
ne
ur
a
l
ne
t
w
or
ks
a
ppr
oa
c
h
f
or
unde
r
s
t
a
ndi
ng
f
r
a
udul
e
nt
f
i
na
nc
i
a
l
r
e
por
t
i
ng,”
I
ndus
t
r
i
al
M
anage
m
e
nt
and D
at
a Sy
s
t
e
m
s
, vol
. 112, no. 2, pp. 224
–
244, 2012,
doi
:
10.1108/
02635571211204272.
[
1
9
]
T
.
R
.
I
z
z
a
l
q
u
r
ny
,
B
.
S
u
b
r
o
t
o
,
a
n
d
A
.
G
ho
f
a
r
,
“
R
e
l
a
t
i
o
n
s
h
i
p
b
e
t
w
e
e
n
f
i
na
n
c
i
a
l
r
a
t
i
o
a
n
d
f
i
na
n
c
i
a
l
s
t
a
t
e
m
e
n
t
f
r
a
u
d
r
i
s
k
m
o
de
r
a
t
e
d
by
a
u
d
i
t
or
q
u
a
l
i
t
y
,”
I
n
t
e
r
na
t
i
o
n
al
J
ou
r
na
l
o
f
R
e
s
e
a
r
c
h
i
n
B
u
s
i
n
e
s
s
a
n
d
S
o
c
i
al
S
c
i
e
n
c
e
,
vo
l
.
8,
no
.
4
,
pp
.
3
4
–
4
3
,
2
0
1
9
,
d
o
i
:
1
0
.
2
05
2
5
/
i
j
r
bs
.
v8
i
4
.
2
81
.
[
20]
J
.
Y
a
o,
Y
.
P
a
n,
S
.
Y
a
ng,
Y
.
C
he
n,
a
nd
Y
.
L
i
,
“
D
e
t
e
c
t
i
ng
f
r
a
udul
e
nt
f
i
na
nc
i
a
l
s
t
a
t
e
m
e
nt
s
f
or
t
he
s
us
t
a
i
na
bl
e
de
ve
l
opm
e
nt
of
t
h
e
s
oc
i
o
-
e
c
onom
y i
n C
hi
na
:
a
m
ul
t
i
-
a
na
l
yt
i
c
a
ppr
oa
c
h,”
Sus
t
ai
nabi
l
i
t
y
, vol
. 11, no
. 6, 2019, doi
:
10.3390/
s
u11061579.
[
21]
K
. R
a
ndha
w
a
, C
. K
. L
oo, M
. S
e
e
r
a
, C
. P
. L
i
m
, a
nd A
. K
. N
a
ndi
, “
C
r
e
di
t
c
a
r
d f
r
a
ud de
t
e
c
t
i
on us
i
ng a
da
B
oos
t
a
nd m
a
j
or
i
t
y vot
i
ng
,”
I
E
E
E
A
c
c
e
s
s
, vol
. 6, pp. 14277
–
14284, 2018, doi
:
10.1109/
A
C
C
E
S
S
.2018.2806420.
[
22]
M
.
N
.
H
oa
ng,
H
.
T
.
L
.
N
guye
n,
a
nd
H
.
N
.
V
i
e
t
,
“
A
m
ode
l
f
o
r
de
t
e
c
t
i
ng
a
c
c
ount
i
n
g
f
r
a
uds
by
us
i
ng
m
a
c
hi
ne
l
e
a
r
ni
ng,”
i
n
T
he
A
nnual
H
aw
ai
i
I
nt
e
r
nat
i
onal
C
onf
e
r
e
nc
e
on Sy
s
t
e
m
S
c
i
e
nc
e
s
, 2022, vol
. 2022, pp. 155
2
–
1561, doi
:
10.24251/
hi
c
s
s
.2022.193.
[
23]
G
.
S
.
T
e
m
pone
r
a
s
,
S
.
A
.
N
.
A
l
e
xa
ndr
opoul
os
,
S
.
B
.
K
ot
s
i
a
nt
i
s
,
a
nd
M
.
N
.
V
r
a
ha
t
i
s
,
“
F
i
na
nc
i
a
l
f
r
a
udul
e
nt
s
t
a
t
e
m
e
nt
s
d
e
t
e
c
t
i
on
t
hr
ough
a
de
e
p
de
ns
e
a
r
t
i
f
i
c
i
a
l
ne
ur
a
l
ne
t
w
or
k,”
i
n
10t
h
I
nt
e
r
nat
i
onal
C
onf
e
r
e
nc
e
on
I
nf
or
m
at
i
on,
I
nt
e
l
l
i
ge
nc
e
,
S
y
s
t
e
m
s
and
A
ppl
i
c
at
i
ons
, I
I
SA
2019
, 2019, pp. 1
–
5, doi
:
10.1109/
I
I
S
A
.2019.8900741.
[
24]
P
.
C
r
a
j
a
,
A
.
K
i
m
,
a
nd
S
.
L
e
s
s
m
a
nn,
“
D
e
e
p
l
e
a
r
ni
ng
f
or
de
t
e
c
t
i
ng
f
i
na
nc
i
a
l
s
t
a
t
e
m
e
nt
f
r
a
ud,”
D
e
c
i
s
i
on
Suppor
t
Sy
s
t
e
m
s
,
vol
.
139
,
2020, doi
:
10.1016/
j
.ds
s
.2020.113421.
[
25]
N
. V
. C
ha
w
l
a
, K
. W
. B
ow
ye
r
, L
. O
. H
a
l
l
,
a
nd W
. P
.
K
e
ge
l
m
e
ye
r
, “
S
M
O
T
E
:
S
y
nt
he
t
i
c
m
i
nor
i
t
y
ove
r
-
s
a
m
pl
i
ng t
e
c
hni
que
,”
J
our
nal
of
A
r
t
i
f
i
c
i
al
I
nt
e
l
l
i
ge
nc
e
R
e
s
e
ar
c
h
, vol
. 16, pp. 321
–
357, 2002, doi
:
10.1613/
j
a
i
r
.953.
[
26]
D
. E
l
r
e
e
dy a
nd A
. F
. A
t
i
ya
, “
A
c
om
pr
e
he
ns
i
ve
a
na
l
ys
i
s
of
s
ynt
he
t
i
c
m
i
nor
i
t
y o
ve
r
s
a
m
pl
i
ng t
e
c
hni
que
(
S
M
O
T
E
)
f
o
r
ha
ndl
i
ng c
l
a
s
s
i
m
ba
l
a
nc
e
,”
I
nf
or
m
at
i
on Sc
i
e
nc
e
s
, vol
. 505, pp. 32
–
64, 2019, doi
:
10.1016/
j
.i
ns
.2019.07.070.
[
27]
I
. G
oodf
e
l
l
ow
, Y
. B
e
ngi
o, a
nd A
. C
our
vi
l
l
e
,
D
e
e
p l
e
ar
ni
ng
,
C
a
m
br
i
dge
,
M
a
s
s
a
c
hus
e
t
t
s
:
M
I
T
P
r
e
s
s
,
2016.
[
28]
D
.
P
.
K
i
ngm
a
a
nd
J
.
B
a
,
“
A
da
m
:
a
m
e
t
hod
f
o
r
s
t
oc
ha
s
t
i
c
opt
i
m
i
z
a
t
i
on,”
ar
X
i
v
-
C
om
put
e
r
Sc
i
e
nc
e
,
pp.
1
-
15,
2017
,
doi
:
10.48550/
a
r
X
i
v.1412.6980
.
[
29]
K
.
H
e
,
X
.
Z
ha
ng,
S
.
R
e
n,
a
nd
J
.
S
un,
“
D
e
e
p
r
e
s
i
dua
l
l
e
a
r
ni
ng
f
or
i
m
a
ge
r
e
c
ogni
t
i
on,”
i
n
P
r
oc
e
e
di
ng
s
of
t
h
e
I
E
E
E
C
om
put
e
r
Soc
i
e
t
y
C
onf
e
r
e
nc
e
on C
om
put
e
r
V
i
s
i
on and P
at
t
e
r
n R
e
c
ogni
t
i
on
, 2016, pp. 770
–
778, doi
:
10.1109/
C
V
P
R
.2016.90.
[
30]
T
.
F
a
w
c
e
t
t
,
“
A
n
i
nt
r
oduc
t
i
on
t
o
R
O
C
a
na
l
y
s
i
s
,”
P
at
t
e
r
n
R
e
c
ogni
t
i
on
L
e
t
t
e
r
s
,
vol
.
27,
no.
8,
pp.
861
–
874,
2006,
doi
:
10.1016/
j
.pa
t
r
e
c
.2005.10.010.
[
31]
N
.
J
a
pkow
i
c
z
,
“
A
s
s
e
s
s
m
e
nt
m
e
t
r
i
c
s
f
or
i
m
ba
l
a
nc
e
d
l
e
a
r
ni
ng,”
I
m
bal
anc
e
d
L
e
a
r
ni
ng:
F
oundat
i
ons
,
A
l
gor
i
t
hm
s
,
and
A
ppl
i
c
at
i
ons
,
pp. 187
–
206, 2013, doi
:
10.1002/
9781118646106.c
h8.
B
I
O
G
R
A
P
H
I
E
S
O
F
A
U
T
H
O
R
S
Y
osua
Efraim
Young
is
a
graduate
candidate
of
Informatics
Gra
duate
Program
from
Universitas
Pelita
Harapan.
He
earned
his
Bachelor’s
Degr
ee
in
Accounting
from
Universitas
Pelita
Harapan
in
Indonesia.
He
can
be
contacted
at
email
:
youngyosua911@
gmail.com.
Hendra
Tjahyadi
is
an
Associate
Professor
of
Informatics
Stu
dy
Program
in
Universitas
Pelita
Harapan
.
He
earned
his
Bachelor’s
Degree
in
Ele
ctrical
Engineering
from
Universitas
Kristen
Maranatha,
Master’s
Degree
in
Instrumentation
a
nd
Control
from
Institut
Teknologi
Bandung,
and
Ph.
D.
in
Control
Engineering
from
School
of
Engineering,
Flinders
University.
His
research
interests
are
in
adaptive
control,
signal
p
rocessing
,
and
artificial
intelligence
. He can be contac
ted at email: hendra.tjahyadi@uph.edu
.
Evaluation Warning : The document was created with Spire.PDF for Python.